MGMT 1040 Lecture Notes - Lecture 22: Scalar Processor, Superscalar Processor, Instruction Set
MGMT 1040 Lecture 22 Notes – Execute Cycle
Introduction
• An instruction fetch-execute cycle that requires six clock cycles from start to finish will
require six clock cycles whether instructions are performed one at a time or pipelined in
parallel with a dozen other instructions.
• It is the average instruction cycle time that is improved by performing some form of
parallel execution.
• If an individual instruction must be completed for any reason before another can be
executed, the CPU must stall for the full cycle time of the first instruction.
• The difference between scalar and superscalar processing with pipelining in the
execution unit.
• In the illustration the execution phase of the fetch-execute cycle is divided into three
parts that can be executed separately.
• Thus, the diagram is divided into steps that fetch, decode, execute, and write back the
results of the execute operation.
• Presumably, each step is carried out by a separate component within the execution unit.
• To simplify the illustration, we have also assumed that in each case the pipeline is full.
• Generally, a single fetch unit pipeline is sufficient to fetch multiple instructions, even
when multiple execution units are present.
• In the scalar processor, each step is assumed to take one clock cycle.
• If the instructions are all of the same length, they will finish consecutively, as shown in
the diagram.
• More complexity in the instruction set will create bubbles in the pipeline.
• Does not alter the basic idea that we are illustrating.
• Panel b of the figure assumes the presence of two execution units.
• It also assumes that the instructions executing in parallel are independent of each other
• That is, the execution of one does not depend upon results from the other.
find more resources at oneclass.com
find more resources at oneclass.com